A new study by scientists from Carnegie Mellon University, the Massachusetts Institute of Technology (MIT), the University of Oxford, and the University of California, Los Angeles, has issued a notable warning about the impact that artificial intelligence (AI) will have on our ability to think.
Research shows that using AI to solve a problem for as little as 10 minutes reduces users’ ability to reason independently, making them more likely to give up when the AI runs out of help.
As AI tools such as ChatGPT, Claude, and Gemini grow in popularity, this study raises the big question: Are humans gradually becoming overly dependent on AI?
The research team conducted three experiments with several hundred participants through an online platform. Participants were asked to solve different types of problems, including basic math and reading comprehension.
Some people will be provided with an AI assistant that can solve tasks directly. However, when the AI tool is suddenly removed from the testing process, this group is more likely to finish faster or give incorrect answers than a group that has never used AI support.
The results were published in a study titled “AI assistants reduce persistence and impact independent performance” on arXiv (operated by Cornell University and considered one of the world’s largest research publications in the fields of technology and science).
Scientists have realized that the problem does not lie in the use of AI in general, but primarily when users allow AI to completely solve a task.
On the other hand, using AI to suggest, explain, and support direction does not cause a significant reduction in thinking ability.
Michelle Bakker, an assistant professor at the Massachusetts Institute of Technology and one of the study’s authors, believes that while AI can clearly help improve productivity in the short term, humans need to be careful when using the technology.
He emphasized that the goal is not to ban AI in learning or work, but to find ways to leverage AI in a way that preserves human learning and problem-solving abilities.
According to Michelle Bakker, the willingness to persist in the face of challenges is a crucial element in helping people acquire new skills and develop their thinking over time. The researchers fear that many people will quickly lose motivation if AI support is removed.
The MIT researchers also believe that future AI systems should be designed for “support and guidance” rather than just providing immediate answers.
This allows users to take advantage of the benefits of AI and maintain their ability to think independently.
The study also warns that AI’s impact on human cognition may occur silently and accumulate over time, especially as AI increasingly appears in learning, work, and daily life.
